I just got my 970. Gigabyte GV-N970WF30C0-4GD.
I was getting 8.8-9kh/s in Windows 7 with:
--algo=scrypt-jane:14 -L 4 -i 1 -l t64x4 -C 1
I haven't tried higher lookup gaps yet.
Which miner you using pls?
I've only managed to get 6.5kh/s so far.
https://github.com/cbuchner1/CudaMiner.gitIf you can compile it yourself, I strongly urge doing it. There are several dependencies that you need, but it really isn't that bad. I added compute50/sm50 support as well for Maxwell cards. It is built with Visual Studio 2010 against the Cuda 6.0 library. If you'd rather have a pre-compiled binary here's what I'm using on my Windows machines:
http://utc.tumblingblock.com/cudaminer_cm10_50.zipMD5Sums
6614a08faec6234fd20c8361d8c080e9 cudaminer.exe
2a4a6c3440f3ec27ea7b2ec417808445 cudart32_60.dll
7812f0f73eda837e9353b3a433abc9a9 pthreadVC2.dll
any good settings you can share for a 750ti?
I do, and I'll share how I found them as well as it will need to be done again with each NFactor change.
The lookup gap is probably one of the most important settings when mining with algos that take a lot of memory. It is essentially a memory to CPU tradeoff. At higher lookup gaps you end up regenerating scratch spaces at the expense of more CPU time, but it ends up being more efficient overall. You can get a 15-33% increase in output by adding the lookup gap. I've found a lookup gap of 4 to work best at NF14 with most of my NVidia hardware. (480, 670, 750ti, 970). Every card is different and they also effect each other as a whole too, so configuring each card with its individual autotune settings may help increase the overall output.
Start off with autotune and try lookup gaps of 4. Add the debug (-D) flag so you can see the progress of the autotune.
-a scrypt-jane:14 -D -l auto -L 4
It will take awhile to grind through the different settings. Pay attention to the final configuration settings that comes up at the end of the autotune. Grab a copy of the console screen with the autotune settings for each GPU, order them by their card number (GPU #
X) in your configuration screen or simply use the most common setting of the group for all of them. And don't be satisfied there. Try the next lookup gaps above and below too until you find that sweet spot. I went through lookup gaps of 1-8 trying to find the best settings.
Autotune Results[2014-10-30 06:26:48] GPU #1: 3578.74 hash/s with configuration
t35x4 <========== Copy each of these. Order by GPU number
[2014-10-30 06:26:48] GPU #1: using launch configuration t35x4
[2014-10-30 06:26:48] GPU #1: GeForce GTX 750 Ti, 422.64 khash/s
[2014-10-30 06:26:48] Temp: 51C Fan speed: 44% Power: N/A
[2014-10-30 06:26:48] 241:2.901| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | kH/s
[2014-10-30 06:26:48] 233:2.787| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | kH/s
[2014-10-30 06:26:48] 238:2.867| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | kH/s
[2014-10-30 06:26:48] 245:2.919| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | kH/s
[2014-10-30 06:26:48] GPU #3: 3575.32 hash/s with configuration
t70x2[2014-10-30 06:26:48] GPU #3: using launch configuration t70x2
[2014-10-30 06:26:48] GPU #3: GeForce GTX 750 Ti, 213.05 khash/s
[2014-10-30 06:26:48] Temp: 54C Fan speed: 46% Power: N/A
[2014-10-30 06:26:49] 242:2.901| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | kH/s
[2014-10-30 06:26:49] 234:2.796| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | kH/s
[2014-10-30 06:26:49] 239:2.864| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | kH/s
[2014-10-30 06:26:49] 243:2.911| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | kH/s
[2014-10-30 06:26:49] 235:2.803| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | kH/s
[2014-10-30 06:26:50] 240:2.876| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | kH/s
[2014-10-30 06:26:50] 244:2.918| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | kH/s
[2014-10-30 06:26:50] 236:2.798| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | kH/s
[2014-10-30 06:26:50] 241:2.878| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | kH/s
[2014-10-30 06:26:51] 245:2.926| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | kH/s
[2014-10-30 06:26:51] GPU #5: 3563.04 hash/s with configuration
t70x2[2014-10-30 06:26:51] GPU #5: using launch configuration t70x2
[2014-10-30 06:26:51] GPU #5: GeForce GTX 750 Ti, 948.35 khash/s
[2014-10-30 06:26:51] Temp: 46C Fan speed: 42% Power: N/A
[2014-10-30 06:26:51] 237:2.813| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | kH/s
[2014-10-30 06:26:51] 242:2.901| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | kH/s
[2014-10-30 06:26:51] 238:2.823| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | kH/s
[2014-10-30 06:26:52] 243:2.895| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | kH/s
[2014-10-30 06:26:52] 239:2.832| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | kH/s
[2014-10-30 06:26:52] 244:2.902| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | kH/s
[2014-10-30 06:26:53] 240:2.843| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | kH/s
[2014-10-30 06:26:53] 245:2.912| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | kH/s
[2014-10-30 06:26:53] GPU #2: 3542.71 hash/s with configuration
t70x2[2014-10-30 06:26:53] GPU #2: using launch configuration t70x2
[2014-10-30 06:26:53] GPU #2: GeForce GTX 750 Ti, 938.02 khash/s
[2014-10-30 06:26:53] Temp: 53C Fan speed: 45% Power: N/A
[2014-10-30 06:26:53] DEBUG: hash <= target
Hash: 00002c74546ed8e76f85aa75db09d4f0fbed962ecb5823c60443af054d4f573a
Target: 0000ffff00000000000000000000000000000000000000000000000000000000
[2014-10-30 06:26:53] GPU #0: GeForce GTX 750 Ti, 3.46 khash/s
[2014-10-30 06:26:53] Temp: 53C Fan speed: 44% Power: N/A
[2014-10-30 06:26:53] accepted: 1/1 (100.00%), 2526 khash/s (yay!!!)
[2014-10-30 06:26:53] 241:2.848| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | kH/s
[2014-10-30 06:26:54] 242:2.864| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | kH/s
[2014-10-30 06:26:55] 243:2.860| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | kH/s
[2014-10-30 06:26:56] 244:2.870| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | kH/s
[2014-10-30 06:26:56] 245:2.876| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | kH/s
[2014-10-30 06:26:56] GPU #4: 3513.91 hash/s with configuration
t70x2ConfigurationHere are the configs for both my 750ti rigs. I'm using Ubuntu Server 14.04 with Cuda 6.0 library- all stock as I haven't had too much stability with bios hacking higher cpu/mem speeds. From what I've seen Windows rigs underperform by a small margin as there is more overhead in the OS.
-a scrypt-jane:14 -D -L 4 -l t35x4,t34x4,t70x2,t70x2,t70x2,t70x2 -o stratum+tcp://utc.tumblingblock.com:3333
[2014-10-30 06:30:40] Stratum detected new block
[2014-10-30 06:30:40] GPU #1: GeForce GTX 750 Ti, 3.40 khash/s
[2014-10-30 06:30:40] Temp: 56C Fan speed: 46% Power: N/A
[2014-10-30 06:30:40] GPU #4: GeForce GTX 750 Ti, 3.33 khash/s
[2014-10-30 06:30:40] Temp: 58C Fan speed: 49% Power: N/A
[2014-10-30 06:30:40] GPU #2: GeForce GTX 750 Ti, 3.35 khash/s
[2014-10-30 06:30:40] Temp: 58C Fan speed: 48% Power: N/A
[2014-10-30 06:30:40] GPU #3: GeForce GTX 750 Ti, 3.28 khash/s
[2014-10-30 06:30:40] Temp: 59C Fan speed: 49% Power: N/A
[2014-10-30 06:30:40] GPU #0: GeForce GTX 750 Ti, 3.46 khash/s
[2014-10-30 06:30:40] Temp: 56C Fan speed: 46% Power: N/A
[2014-10-30 06:30:40] GPU #5: GeForce GTX 750 Ti, 3.34 khash/s
[2014-10-30 06:30:40] Temp: 50C Fan speed: 43% Power: N/A
[2014-10-30 06:30:55] DEBUG: hash <= target
Hash: 000001ad7b2379a4fd4d5e0f8b699ae52efb4c1251f3895c1aa500f62db5b2ef
Target: 0000247a2f949c7b000000000000000000000000000000000000000000000000
[2014-10-30 06:30:55] GPU #0: GeForce GTX 750 Ti, 3.46 khash/s
[2014-10-30 06:30:55] Temp: 56C Fan speed: 46% Power: N/A
[2014-10-30 06:30:56] accepted: 30/30 (100.00%), 20.17 khash/s (yay!!!)
On another 6x 750ti rig I ended up using the autotune settings at a lookup-gap of 4 and defined the kernel settings for each card as it seemed to help out the last card which was lagging behind the others a bit.
-a scrypt-jane:14 -D -L 4 -l t35x4,t35x4,t35x4,t35x4,t35x4,t70x2 -o stratum+tcp://utc.tumblingblock.com:3333
[2014-10-30 06:31:16] GPU #4: GeForce GTX 750 Ti, 3.27 khash/s
[2014-10-30 06:31:16] Temp: 61C Fan speed: 51% Power: N/A
[2014-10-30 06:31:16] GPU #1: GeForce GTX 750 Ti, 3.34 khash/s
[2014-10-30 06:31:16] Temp: 58C Fan speed: 46% Power: N/A
[2014-10-30 06:31:16] GPU #3: GeForce GTX 750 Ti, 3.36 khash/s
[2014-10-30 06:31:16] Temp: 56C Fan speed: 46% Power: N/A
[2014-10-30 06:31:16] GPU #2: GeForce GTX 750 Ti, 3.33 khash/s
[2014-10-30 06:31:16] Temp: 56C Fan speed: 46% Power: N/A
[2014-10-30 06:31:16] GPU #5: GeForce GTX 750 Ti, 2.89 khash/s
[2014-10-30 06:31:16] Temp: 56C Fan speed: 47% Power: N/A
[2014-10-30 06:31:16] GPU #0: GeForce GTX 750 Ti, 3.53 khash/s
[2014-10-30 06:31:16] Temp: 57C Fan speed: 46% Power: N/A
[2014-10-30 06:31:20] GPU #1: GeForce GTX 750 Ti, 3.42 khash/s
[2014-10-30 06:31:20] Temp: 57C Fan speed: 46% Power: N/A
[2014-10-30 06:31:20] accepted: 25624/25734 (99.57%), 19.80 khash/s (yay!!!)
For some lulz I had my 670 at 3.8khash/s with some overclocking- also with a lookup gap of 4. It actually outperformed the 750ti at stock speeds, but with around 3x as much wattage. I couldn't quite get it to 4khash/s by upping the vcore voltage and cpu and memory speeds. My 970 on windows gets around 9khash/sec with a lookup gap of 4.